Wu Yi

The user can choose to use different base images to adapt his model definition, e.g. tf-2.0, tf-1.15, pytorch-1.0 etc. SQLFlow provides a list of base images for users to choose,...

 SQFlow is a end2end ML tool set  SQLFlow is a bridge between DBMS and AI

Current errors in the experiment: https://github.com/sql-machine-learning/sqlflow/pull/2618, we need to find out the reason and solution. ``` === RUN TestFeatureDerivationNoColumnClause coverage: 25.5% of statements panic: test timed out after 2m0s goroutine...

Well, in my mind, proposal 3 is quite the same as the proposal 1, because: - in proposal 1, the Go code must know the actual Python API details and...

> In the proposed architecture, the IR is not "Go side", it's a formal protocol between the Parser (Go side) and the Runtime (Python side), and that's why #2335 prefers...

Agree with this design so that script contributor is not writing functions from scratch, instead, they can just write: ```python def user_run_function(gen, args, ...): for row in gen: # deal...

I'd like to have the all-in-one Docker image back. For user quick start purpose only. And the current "local mode" is easy to maintain. For using custom model images like...

What is the value of `SQLFLOW_DATASOURCE` did you set?

Please try `mysql://sqlflow:sqlflow%123@tcp(10.11.25.13:3306)/?`

Do we have to apply autoML with auto feature derivation? e.g. if we add `STANDARDIZE` to all numerical columns, is it affecting the model performance respectively?